Aug 30, 1999 - We have determined several of the basic thermodynamic properties of eglin c ... To move the gene III fusion point further away from the binding ..... enrichment we included a teglin/m663 mixture and a mixture of m663 ..... What else have we learned about making protein based inhibitors and stromelysin? 1.
»
4
AD
Award Number DAMD17-94-J-4270
TITLE: A New Generic Method for the Production of Protein-Based Inhibitors of Proteins Involved in Cancer Metastasis
PRINCIPAL INVESTIGATOR: Marshall H. Edgell, Ph.D.
CONTRACTING ORGANIZATION: University of North Carolina, Chapel Hill Chapel Hill, North Carolina 27599-4100
REPORT DATE: August 1999
TYPE OF REPORT: Final
PREPARED FOR:
U.S. Army Medical Research and Materiel Command Fort Detrick, Maryland 21702-5012
DISTRIBUTION STATEMENT: Approved for Public Release; Distribution Unlimited
The views, opinions and/or findings contained in this report are those of the author(s) and should not be construed as an official Department of the Army position, policy or decision unless so designated by other documentation.
20000426 096 DTIC QUALITY INSPECTED 3
Form Approved OMB No. 0704-0188
REPORT DOCUMENTATION PAGE
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this' collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 121B Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project 10704-0188), Washington, DC 20503
1. AGENCY USE ONLY (Leave blank)
2. REPORT DATE
3. REPORT TYPE AND DATES COVERED
August 1999
Final (lAug 94-31 Jul 99)
4. TITLE AND SUBTITLE
5. FUNDING NUMBERS
A New Generic Method for the Production of Protein-Based Inhibitors of Proteins Involved in Cancer Metastasis
DAMD17-94-J-4270
6. AUTHOR(S)
Marshall H. Edgell, Ph.D.
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
8. PERFORMING ORGANIZATION REPORT NUMBER
University of North Carolina, Chapel Hill Chapel Hill, North Carolina 27599-4100
9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES)
10. SPONSORING / MONITORING AGENCY REPORT NUMBER
U.S. Army Medical Research and Materiel Command Fort Detrick, Maryland 21702-5012
11. SUPPLEMENTARY NOTES
12a. DISTRIBUTION/AVAILABILITY STATEMENT
12b. DISTRIBUTION CODE
Approved for Public Release; Distribution Unlimited
13. ABSTRACT (Max/mum 200 words)
Our objective was to develop a method to make protein-based inhibitors against protein targets and as the test case to a proteinase involved in metastasis, stromelysin. The approach used phage display and a display framework which was expected to bind preferentially to the active site pocket of target proteins. While we were able to find peptides in phage display libraries that bound to active stromelysin or to cadmium inactivated stromelysin and could find binders to a test target, papain, in a constrained loop library we were unable to get the fully constrained scaffold protein to bind to its cognate target, subtilisin in the phage display system. Hence the overall objective was not obtained. As part of our efforts to characterize the binding epitope display framework, eglin c, we did develop a new method for using mutagenesis to study protein structure which we call patterned library analysis. We carried out a proof-of-principle experiment using the new method in which we were able to reproduce the values for cc-helix propensity indicating that the method does indeed work. This method should have broad utility as a method for the assessment of hypotheses concerning the determinants of protein structure.
14. SUBJECT TERMS
15. NUMBER OF PAGES
Breast Cancer
94 stromelysin, phage display, eglin c, combinatorial libraries
17. SECURITY CLASSIFICATION OF REPORT
Unclassified NSN 7540-01-280-5500
18. SECURITY CLASSIFICATION OF THIS PAGE
Unclassified
16. PRICE CODE
19. SECURITY CLASSIFICATION OF ABSTRACT
20. LIMITATION OF ABSTRACT
Unclassified
Unlimited
Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. Z39-18 298-102
USAPPC V1.00
FOREWORD
Opinions, interpretations, conclusions and recommendations are those of the author and are not necessarily endorsed by the U.S. Army. Where copyrighted material is quoted, permission has been obtained to use such material. Where material from documents designated for limited distribution is quoted, permission has been obtained to use the material. Citations of commercial organizations and trade names in this report do not constitute an official Department of Army endorsement or approval of the products or services of these organizations.
mA&T
In conducting research using animals, the investigator(s) adhered to the "Guide for the Care and Use of Laboratory Animals," prepared by the Committee on Care and use of Laboratory Animals of the Institute of Laboratory Resources, national Research Council (NIH Publication No. 86-23, Revised 1985). For the protection of human subjects, the investigator(s) adhered to policies of applicable Federal Law 45 CFR 46. In conducting research utilizing recombinant DNA technology, the investigator(s) adhered to current guidelines promulgated by the National Institutes of Health. In the conduct of research utilizing recombinant DNA, the investigator(s) adhered to the NIH Guidelines for Research Involving Recombinant DNA Molecules. In the conduct of research involving hazardous organisms, the investigator(s) adhered to the CDC-NIH Guide for Biosafety in Microbiological and Biomedical Laboratories.
PI - Signature \
Date
Table of Contents
Page Front Cover SF 298 Report Documentation Foreword Table of Contents Introduction Body Summary Eglin as a Phage Display Scaffold Truncated Eglin as a Scaffold Exploring Search Strategies for Binders Using Randomized Peptides Production of the Target Protein: Stromelysin Characterization of the Scaffold Protein: Eglin C Key Research Accomplishments Reportable Outcomes Conclusions Personnel Supported
1 2 3 4 6 7 7 9 13 15 16 19 22 23 24 25
Figure 1. Figure 2. Figure 3 Figure 4 Figure 5 Figure 6 Figure 7a Figure 7b Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13
26 27 28 29 30 31 32 33 34 35 36 37 38 39
Ribbon Diagram of Eglin C. E coli lysates expressing pegin or eglin c Peglin Binding to Wells Coated with Subtilisin Earl strategy for library construction. Construction of a 'White' mB AX Phage. Inhibitory Effects of Various Phage Preparations without Dilution Binding (or lack of binding) of Peglin to Stromelysin Binding (or lack of binding) of Peglin to Stromelysin Conversion of Egin to Teglin. Conversion of Tegin to Teglin-papain Construction of a Teglin Based Papain Cleavge Sequence Library ELISA Values for Clones After Panning Against Papain. Sequences of Binders to Papain ELISA Values for Clones After Panning Against Stromelysin
Page 4
Figure 14 Figure 15 Figure 16
Clones After Panning Against Stromelysin with MgCl2 Clones After Panning Against Stromelysin with CdCl2 Sequences of Clones Which Bound to CdCl2 Treated Stromelysin
40 41 42
Figure 17 Figure 18 Figure 19 Figure 20 Figure 21 Figure 22 Figure 23 Figure 24 Figure 25
Alignments of Sequences from Stromelysin Binding Clones Standard Curve for Stromelysin. Trypsin Activation of Prostromelysin Gel Assay for Trypsin Activation of Prostromelysin Trypsin Activation of Prostromelysin at 4.16 uM Trypsin Crude Lysate Activation of Prostromelysin Made by Our Clones Induction of the Prostromelysin Strom9 Clone in E.coli Fractions from Column Purification of His-Tagged Prostromelysin Quantitation of His-tagged Prostromelysin Yield from Column
43 44 45 46 47 48 49 50 51
Table 1 Table 2 Table 3 Table 4
Binding of Circularly Permuted Eglin (Peglin) to Subtilisin Phage Displaying Teglin Variants Binding to Papain Enrichment Factors for Isolates from the Teglin-Papain Library Binders to the General Enzyme Panel
52 53 54 55
Appendix. Manuscripts: Published and Submitted 56 1. Waldner, JC, Lahr, SJ, Edgell, MH, and Pielak, GJ. Effect of a polyhistidine terminal extension on eglin c stability. Anal. Biochem. 263 116-118 (1998) 2. Waldner, JC, Lahr, SJ, Edgell, MH, and Pielak, GJ. Nonideality and protein thermal denuration. Biopolymers 49 471-479 (1999) 3. Lahr, SJ, Broadwater, A, Carter, CW, Jr., Collier, ML, Hensley, L, Waldner, JC, Pielak, GJ, and Edgell, MH. Patterned library analysis: a method for the quantitative assessment of hypotheses concerning the determinants of protein structure. PNAS (1999)
Page 5
INTRODUCTION Our long range interests were to design an array of protein-based inhibitors for proteins involved in various steps in metastasis. The primary objective of this project was to test the viability of a new approach for designing inhibitors for proteins and as 'proof-of-concept' to generate an inhibitor for stromelysin, a proteinase implicated in metastasis. Our approach was to use protein engineering to redirect the activity of a naturally occurring proteinase inhibitor, eglin c to the targets of interest. Since the state-of-the-art of protein structure/function is such that it would be unlikely that we could design an effective inhibitor in a single pass, we planned to use genetic screening techniques to find the highest affinity variants in large libraries of structural variants centered around the basic design concept. Our approach used phage display for genetic screening and a display framework which was expected to bind preferentially to the active site pocket of target proteins and to provide the constraints necessary for high-affinity binding. Hence the project consisted of: 1. 2. 3. 4. 5. 6.
getting the phage display system working in our laboratory. producing stromelysin for use as a binding target getting our protein scaffold to work in phage display determining how to get binding to proteinases that can cleave the binding molecules building suitable libraries for binding to stromelysin finding binders to stromelysin
We were able to accomplish steps 1,2,4,5 and 6 but not 3. We were able to get all of the parts of the project to work except the crucial one involving the scaffold protein that was intended to provide the constraints necessary for high-affinity binding. Early in the project, before we had become familiar with some of the eccentricities of the phage display system, we obtained evidence that we interpreted as positive concerning the utility of our scaffold protein. Late in the project we discovered that the scaffold protein did not, in fact, provide the display function necessary for phage display. This made it impossible for us to generate high-affinity inhibitors using the approach envisioned in this project. We are disappointed that this project has not made a direct contribution to the attack on breast cancer as was the project goal. However, as part of our efforts to characterize the binding epitope display framework protein, eglin c, we did develop a new method for using mutagenesis to study protein structure which we call patterned library analysis. We carried out a proof-of-principle experiment using the new method in which we were able to reproduce known values for a-helix propensity indicating that the method does indeed work. We expect the method to be used in various approaches to exploring the determinants of protein structure. This should make a contribution to protein structure prediction, one of the outstanding unsolved problems in biology. Like most basic science tools, that should, in the long term, enhance our capacity to deal more effectively with diseases such as cancer.
Page 6
BODY Summary Initial Project Goals 1. Convert eglin c into a framework for the development of protein-based protein inhibitors. 2. Incorporate a binding epitope for stromelysin into the inhibitor framework and identify stromelysin binders in libraries designed to optimize binding to the desired target. 3. Test the stromelysin binding eglin variant for its capacity to inhibit metastasis. Failures 1. The main project failed ultimately in step 1 although we did get some stromelysin binders by using simplified scaffolds (step 2). a. During the first year we discovered that wild type eglin did not work in the phage display system which was central to our project. We assumed that this was due to the site on eglin to which it was fused to the phage attachment protein, pill. b. Consequently we constructed a circularly permuted version of eglin variant (peglin) to move the Nterminus to the opposite side of eglin to overcome this problem. c. Control experiments indicated that this modified form of eglin did work in the phage display system. d. A large number of experiments were then done to improve our use of the phage display system, to enhance library construction, to make stromelysin, and to document that binders could be found to target proteinases. We found binders to various enzymes using randomized peptides and to stromelysin using a simplified scaffold consisting of an 18 amino acid loop. e. When we began to use peglin under phage display conditions that were now known to be reliable we were not able to get peglin to bind to its cognate enzyme. f. After considerable frustration we carried out a reassessment at the beginning of the last year of the project as to whether peglin would function at all as a phage display scaffold and discovered that it did not! As this was very near the end of the grant period we did not worked out the reason for this problem although we understand the source of the difference in our current results and the earlier ones. Instead we used the remaining time to explore a successful ancillary project that had arisen during the main project. Accomplishments 1. We carried out 'proof-of-principle' experiments for a new method to utilize mutagenesis and combinatorial libraries to assess hypotheses concerning the determinants of protein structure. This project grew out of our efforts to characterize the model protein that we have been using as a scaffold. a. As proof of principle we used our new method to show that it could reproduce known values for the helix propensities of amino acids (ref. 3). b. We feel that the development of this new method is a very significant one in that we expect it will be used broadly to enhance protein structure prediction algorithms. Protein structure prediction is, of course, one of the major outstanding problems in biology.
Page 7
2. Characterization of eglin c. a. We have determined several of the basic thermodynamic properties of eglin c and shown that these apply to his-tagged eglin c as well as to wild type (ref. 1). b. We have uncovered a non-ideality between the native and denatured states of eglin c that explains the differences between van't Hoff and calorimetric denaturation enthalpies. This observation may, in fact, apply to other proteins whose behavior is otherwise consistent with a two-state mechanism for unfolding (ref. 2).
Page 8
Objectives of the Project Our long range interests were to design an array of protein-based inhibitors for proteins involved in various steps in metastasis. The primary objective of this project was to test the viability of a new approach for designing inhibitors for proteins and as 'proof of concept' to generate an inhibitor for stromelysin, a proteinase implicated in metastasis. Our approach was to use protein engineering to redirect the activity of a naturally occurring proteinase inhibitor, eglin c to the targets of interest. To redirect the activity of eglin c the plan was to retain the native constraints in the framework that reduce the number of non-productive conformations accessible to the binding epitope but to replace the native binding epitope with one for stromelysin. Since the state of the art of protein structure/function is such that it would be unlikely that we could design an effective inhibitor in a single pass, we planned to use genetic screening techniques to find the highest affinity variants in large libraries of structural variants centered around the basic design concept. The plan was to employ multiple cycles of design, construction, affinity screening, and biophysical analyses. Once high affinity (Kd > 10~9 M) stromelysin inhibitors were identified we planned to test their effects in standard assays for metastasis. Investigations. Most of the investigations described below were done in parallel. They are reported here as isolated projects to provide a more coherent report. Hence it is important to keep in mind when reading a section that we did not necessarily know at the time we did those experiments everything that has been described in earlier sections of the report. The main conclusion of section I in which we show that the basic premise of our project was flawed, at least in terms of the model protein chosen, was not found out until the last year of the project. We describe how our early results (mid 1996) although correct and reproducible misled us in terms of the properties of our model protein. On August 23, 1998 we collected data that indicated that the scaffold protein chosen was not going to work in the context we were trying. From that point on we focused on a scientifically promising lead that had developed in our laboratory while doing basic characterization of the eglin c scaffold protein. This ancillary project has been successful and we hope that it will provide a new tool to help in the characterization of proteins and the determinants of protein structure. We are disappointed that this project has not made a direct contribution to the attack on breast cancer as was the project goal. However, the successful components of the project have led to a new method with broad basic science implications. We expect the method to be used in various approaches to exploring the determinants of protein structure. This should ultimately enhance our capacity for protein structure prediction. Like most basic science tools, that should, in the long term, accelerate learning about cancer which will ultimately increase our capacity to deal more effectively with the disease. I. Eglin c as a Phage Display Scaffold. Our basic approach was to extend the reach of traditional protein engineering by constructing large libraries of structural variants around a central design concept and then using the phage display system to screen the population for binders. Hence it came as a considerable shock when we discovered that wild-type eglin c, when fused to the Ml 3 gene III protein in the M13 particle, does not bind to a target to which the free inhibitor
Page 9
normally binds (e.g. subtilisin). We presumed that this problem was due to the fact that the eglin c molecule attaches to the phage via it's C-terminus which is close to the active site and that this blocks access to the eglin c binding epitope (see Figure 1). That is, the eglin c C-terminus appears to be too close to the loop containing the residues that bind with the target proteinase. To move the gene III fusion point further away from the binding epitope we made eglin c variants with various C-terminal truncations and made the Ml 3 gene III fusions via linkers of 5 and 9 prolines. None of these eglin c variants bound to subtilisin as measured by enrichment in a phage display assay. Peglin Construction. We then made a circularly permuted version of eglin c in which the C-terminus was moved to the side of the protein opposite from the residues that bind with the target. We designed an eglin variant in which the wild-type N and C termini have been joined together and new termini created by opening up a tight turn on the opposite side of the protein (Figure 1). This involved various modeling exercises to evaluate residues for a tight turn and at exactly which point to truncate the existing N and C termini. A construction was carried out and was verified by sequencing. We then tested the new construct for its ability to inhibit subtilisin and function in phage display. Peglin is Active Against Subtilisin. The peglin construct, in the plasmid pET28, was transformed into E. coli BLR/y.vS and grown up in 2YT. Cell lysates were prepared by resuspending a frozen cell pellet in 50 mM Tris pH 8.0 and treatment with 100 ug/ml of lysozyme. Samples boiled for 10 minutes in 1% SDS were then examined on 15% acrylamide gels (Figure 2). This indicated that the protein was relatively stable since if it were not the case there would be less peglin than eglin in the lysates. Peglin had a specific activity in these lysates essentially the same as eglin; relative specific activity of 1.05 as measured by its capacity to inhibit subtilisin. This tells us that the circularly permuted version of eglin does indeed bind to its normal target, subtilisin. It does not tell us whether peglin will function in the phage display system. Initial Evidence that Peglin Binds in the Phage Display System (~May 1996). The gene for the circularly permuted version of eglin was then transferred to the Ml 3 phage used for phage display. That is, the peglin gene was fused to the N-terminus of the M13 pill gene. To test for binding we used a target proteinase binding assay. Samples are applied to wells in a 96-well plate coated with subtilisin and the amount of phage that binds to the wells is assessed with alkaline phosphatase conjugated anti-M13 antibody. We grew up liter cultures of the Ml 3 phage containing the peglin fusion product and mixed phage with subtilisin coated wells to see if it would bind. To get a decent molar ratio of phage to enzyme we concentrated the phage using polyethylene glycol precipitation. Peglin fusion phage were then applied to subtilisin coated wells and the amount of binding assessed using alkaline phosphatase labeled antibody against Ml 3. Various dilutions of the concentrated phage stock were employed (Figure 3). As a control for this experiment equal numbers of phage expressing a phage display epitope for strepavidin was applied to wells coated with subtilisin (Figure 3). 1.4 times as many peglin phage bound to subtilisin coated wells as did control phage. This weak binding was attributed to the presence of phosphate in the binding media which reduces subtilisin activity. To further assess peglin binding to subtilisin we carried out an enrichment assay. In such an assay we mixed a small quantity of peglin with a large population of non-binding phage and subjected the population to several rounds of panning against subtilisin. One then looks for a change in the input ratio. This can be done Page 10
easily since the peglin phage contains a gene for the alpha fragment of beta-galactosidase and hence will make blue plaques on indicator plates. The non-binding phage do not carry the alpha fragment and hence make white plaques. Wells in a 96-well plate were coated by exposure to solutions of subtilisin at 25 ug/ml and incubation overnight. The wells were washed 5 times with 50 mM Tris pH 8.0 and then treated for 12 hours with 1% BSA to block any remaining non-specific binding sites in the wells. Phage were added to the wells and allowed to bind for 4 hours. The wells were then washed eight times with PBS-0.1% Tween 20 and then the bound phage were eluted with 50 mM glycine-HCl pH 2.0. After neutralization using 200 mM NaP04 pH 7.5 the eluted phage were titered. This assay gave an enrichment of anywhere from 7 to 466 fold (Table 1). These experiments convinced us that peglin could bind to subtilisin, the normal target for the eglin epitope and hence we proceeded with a series of experiments over the next two and a half years to transfer other epitopes to the peglin scaffold. The problems with this interpretation are delineated in the section below on Final Evidence that Peglin Really Does Not Bind to Subtilisin (August 1998). Enhancements to the Peglin Phage Display Vector. I. Restriction Sites. Construction of libraries seemed to take much more time than was convenient for this project that was so dependent on library construction. It seemed to us that the problem was that each library was an idiosyncratic affair involving different restriction sites. Each new restriction site employed seemed to take a large amount of time for us to optimize the conditions for effective cleavage while leaving ligatable ends. We decided that it would be worth the effort to build a vector that would use the same restriction enzymes independent of where we intended to make the mutations in the parent gene. To do this we intended to use a restriction enzyme such as Ear I that cleaves outside of the enzyme recognition sequence. If one uses PCR primers that have such a site at their ends the PCR product can be cleaved to leave a ligatable ends inside of the template sequence at any position of choice independent of any restriction sites in the template (Figure 4). This also means that one can use the same restriction enzyme for any library. Optimization for one library should carry over to all of those that follow. To make this work we needed a vector that had no Ear I sites. Our original peglin containing Ml3 phage derivative had two Ear I sites. These were removed by site directed mutagenesis. Enhancements to the Peglin Phage Display Vector. II. White Plaque Variant. One of the methods to find rare phage which bind to a target of interest is to look for enrichment of the variants containing potential binding epitopes relative to non-binder phage. One way to measure enrichment is to be able to distinguish library phage from non-binder phage based on the color of plaques they make. Our library vectors all made blue plaques on indicator plates. While we had several non-binder phage that gave white plaques on indicator plates they all produced many more phage progeny per generation than did the mBAX phage into which peglin was inserted. We decided to prepare a white plaque variant of the mBAX vector, that is, the base vector without peglin. This was done using PCR primers to amplify up all of the mBAX phage except for a 200 bp deletion within the beta-galactosidase alpha fragment (Figure 5). Libraries Involving the Peglin Scaffold. The first peglin based library was constructed by removing an Xho/Xba fragment from peglin and replacing it with a degenerate PCR product generated from degenerate oligonucleotide primers. This library had eight residues in the center of the binding epitope loop (Figure 1) randomized and replaced two arginines with alanine. The two arginines in the native molecule interact with Page 11
residues in the loop (an aspartic acid and a threonine). Since those loop residues have been randomized we chose to remove the arginines. The diversity of this library was low, about 3 x 105 different phage variants. This library was never tested against stromelysin since at the time we made the library we did not have any stromelysin and by the time we had obtained a useful supply of stromelysin we had learned that the peglin scaffold was not working in the phage display system. Final Evidence that Peglin Really Does Not Bind to Subtilisin (August 1998). Having decided initially that peglin fit our preconceptions and was indeed a suitable scaffold protein for displaying binding epitopes for novel targets we set about learning how to use the phage display system effectively, to working out methods to enhance the library construction process, to making suitable quantities of stromelysin, and to document that binders could be found to target proteinases. Once the phage display system was under control and we began to use peglin under conditions that were now known to be reliable we discovered that could not get peglin to bind to its cognate enzyme. Hence we were forced to reassess the basic premise that peglin was a suitable scaffold for phage display. That meant reexamining whether the unmodified peglin protein with the subtilisin epitope would bind to subtilisin. We retested whether peglin would inhibit subtilisin. One possibility for our lack of success at getting binders was that the peglin strain had mutated and become inactive, so we made several new phage preparations from old isolates. Large cultures were grown up to produce enough phage to be able to detect inhibition. When these polyethylene glycol concentrated stocks were mixed in various dilutions with subtilisin the peglin stock inhibited subtilisin more completely than did the non-binder phage stock just as in the earlier experiments (Figure 3). However, now looking more intensely for a problem with these results, I realized that the control non-binder phage produced a much higher concentration of phage in the growth cultures than the peglin phage and hence when comparing equal quantities of phage (PFU/ml) we were adding much different volumes of the two concentrated phage stocks. The experiment was then repeated using equal volumes of phage stocks instead of equal amounts of phage. In that experiment we discovered that equal volumes of the phage stocks gave exactly the same amount of inhibition (Figure 6) even though the amounts of phage in equal volumes differed by a factor of 100! In particular, look at Figure 6, rows C,D, and E. These different phage preparations behaved exactly the same in terms of inhibition even though the amount of putative inhibitor, that is, phage, were quite different in the preparations. The implication ofthat was that the inhibition was due to a contaminant in the concentrated phage stocks, most likely polyethelene glycol, rather than the phage itself. This was repeated by two different people in the lab and with several different peglin isolates. Another test done to assess peglin binding to subtilisin was to take phage preparations and see whether they bound to subtilisin adhering to wells in 96-well plates. In this case binding was assessed by using alkaline phosphatase conjugated antibody against Ml 3 to measure the amount of phage bound to the well. In this test we compared the amount of phage binding to wells containing subtilisin with wells containing catalase to control for non-specific binding. Testing ten different peglin isolates we found no binding to subtilisin that was higher than the non-specific control (Figure 7). Evidence for binding by one of the peglin isolates would be larger numbers in any given column in rows A-D (the subtilisin coated wells) versus E-H (the catalase coated wells). None of the isolates gave such a result. Page 12
These experiments indicated that peglin did not bind to subtilisin. The other piece of data originally indicating that peglin bound to subtilisin was an enrichment experiment in which blue plaque producing peglin phage were mixed with white plaque producing non-binder control phage and the mixture panned against a subtilisin coated well. As we became more experienced with phage display in the two years since those initial experiments we learned that single round panning experiments were unreliable. They often gave irreproducible results. In addition, the single round competition experiments were very sensitive to the relative burst sizes of the two phage and this added to the problems with a single round panning experiment as an assay for binding. As a consequence we often obtained results from single round panning experiments that did not stand up on reproduction. That was a gradual realization and it did not cause us to question our original conclusions about peglin since we thought we had other convincing data that peglin would bind to subtilisin. These new experiments concerning peglin's ability to bind to subtilisin were done in August of 1998. We concluded at that time from the data from those experiments that peglin was not going to work as a scaffold for creating new protein-based inhibitors. Our laboratory efforts for the remaining 12 months were then turned to a scientifically promising lead that had developed from the ancillary project involving the biophysical characterization of the base scaffold protein, eglin c. That project has been quite successful and is described in section V below. II. Truncated Eglin as a Scaffold. It has been shown by others that a truncated form of eglin inhibits it's normal targets with the same efficiency as does the wild type eglin protein. So, while circular peptides would not make good physiological inhibitors due to their susceptibility to proteolytic degradation they could serve as a scaffold for the discovery of binding epitopes that could then be moved to the full circularly permuted form of eglin, peglin. So, in parallel with our efforts to develop the circularly permuted form of eglin as a scaffold we began to explore the utility of this truncated form of eglin as a scaffold to search for binding epitopes targeted to other proteins. We called the truncated form teglin. Teglin consists of an eighteen amino acid circular peptide closed via a cysteine bond and fused to the pill Ml3 protein for display via a linker of QGGGG (Figure 8 ). We tested teglin for binding to subtilisin by both the inhibition assay and single cycle enrichment. Polyethelene glycol concentrated teglin phage inhibited subtilisin. We got no enrichment from a single cycle. At the time we inteipreted this as evidence that teglin worked and that enrichment failed due to the anticipated very tight binding of teglin to subtilisin. Knowing what we know now I can't say whether or not teglin inhibited subtilisin. As the next section will indicate we do know that teglin does work in phage display as would be expected from experiments that others have done with constrained loops and phage display. Conversion of Teglin to a Papain Binder. A. The Papain Cleavage Epitope. This project was a methods development project. Our ultimate target for inhibitor development was stromelysin, a proteinase implicated in metastasis. However, to use stromelysin as a target we needed to make significant quantities of the protein. That was being pursued in a parallel activity and is described below (section IV). To work out methods we used a commercially available enzyme, papain, as a target proteinase. We chose papain since it is cheap, it's structure is known, it is a proteinase from a different class of enzymes (cysteine proteinase) than Page 13
subtilisin (serine proteinase) and it's cleavage specificity is known. We also wanted to work with a proteinase since there are serious issues to work out for using phage display to find protein binders to a proteinase. One could expect background binding of proteins to a proteinase. One could expect that intermediate level binders might be cleaved by the proteinase and released from the affinity matrix preventing enrichment. The binding epitope in teglin was then replaced by a sequence that is cleaved by papain (Figure 9 ). This construction was verified by sequencing. Two isolates of the modified teglin display phage (white plaque producer) were then mixed with an M13 phage, m663, displaying a peptide for a non-cognate target that produces blue plaques on indicator plates. These mixtures were tested for enrichment in wells of a 96-well plate coated with either active or papain inactivated by treatment with Mg+2. To control for unspecific binding and enrichment we included a teglin/m663 mixture and a mixture of m663 with an Ml3 without any display peptide. We found about a ten fold greater enrichment for the teglin variants with the papain binding epitope relative to the m663 phage displaying a peptide for a non-cognate target (Table 2). On the other hand teglin with the subtilisin binding epitope was enriched 3-5 fold. Conversion of Teglin to a Papain Binder. B. Papain Cleavage Epitope Library. A library of binding epitopes in the teglin scaffold was constructed (Figure 10) by insertion of an Xho/Xba fragment to replace the subtilsin binding epitope. The new fragment maintained the papain cleavage sequence, phe-ala, and randomized four residues, one just upstream of the cleavage site and three downstream. The library was subjected to four rounds of enrichment in wells coated with inactivated papain. Thirty-four phage isolates from the enriched population were then tested by a single round of competition against m663 a blue plaque producer on X-gal indicator plates. Two of the phage isolates had considerable better enrichment than the rest (Table 3). We carried out another set of panning with the library against inactivated papain. This time we used six rounds of panning with each round being followed by amplification to increase the diversity of the phage at each round. We had discovered that only a few thousand phage were getting through the first two rounds of panning prior to amplification. This was nice for getting only a few phage types out of the panning process but not good for trying to find all of the different types of binders. After the panning 90 isolates were picked and tested for binding to inactivated papain using alkaline phosphatase conjugated anti-M13 antibody to assess binding (Figure 11). These panned phage divided up into three groups, 21 non-binders, 11 intermediate binders, and 58 high binders. Eight isolates were sequenced (Figure 12). All five from the high binder group had undergone some mutation that eliminated the cysteine link to form the constrained loop. Only two of the eight retained the cysteine constraint. This result with papain as the binding target had both positive and negative interpretations. On the positive side we had been able to retarget the eglin binding loop to a non-cognate proteinase target. We were able to see binding to a proteinase that might have cleaved the binding epitope preventing enrichment. On the down side most of the binders had lost the cysteine which was designed to hold the binding loop closed. This converted the binding sequence into a free peptide. The trouble with free peptide binders is that there is an upper limit on the binding potential for peptides imposed by the large entropy penalty paid from the many degrees of freedom unbound state to the reduced degrees of freedom bound state. This upper limit for the
Page 14
binding of free peptides is below what is usually thought to be necessary for getting enough binding under physiological conditions to produce an effect. Construction of General Libraries Using the Teglin Scaffold. Despite the downsides, the papain results seemed promising to us and so we constructed two general libraries that we thought could be used to look for stromelysin binders. Initially we had imagined using libraries like the one for papain but which incorporated the stromelysin cleavage sequence. However, results using randomized peptides that will be discussed below suggested to us that less specific libraries might be a more effective use of our time and of course less specific libraries would certainly have more utility for use with different targets. In wild-type eglin there are two interactions between residues in the binding loop and the underlying beta-strands that increase the affinity of binding. We made one library that replaced the arginines in the betastrand with alanines and another library that put randomized codons at those sites. In the latter library the idea was to provide an opportunity for different binding loop/beta-strand interactions to form. In both libraries we randomized eight residues centered on what would be the scissile bond in a cleavable peptide. The first library, 8X+2A, had a complexity of 2.3 x 108. The second library, 8X+2X, had a complexity of 1.5 x 109. These libraries had quite good diversity which we attribute to benefits of using the Ear I strategy and optimized vector described in a section above (Enhancements to the Peglin Phage Display Vector. I. Restriction Sites) which were also applied to the teglin vector. Utilization of Teglin Libraries. We began to use these libraries against various targets, a bank of model targets (cheap, commercially available enzymes), stromelysin and subtilisin. As the binding controls for these experiments we added the circularly permuted version of eglin that is supposed to bind to subtilisin to a catalase binder we had isolated and a strepavidin binder. In contrast to the other binding controls, the peglin control did bind to its cognate target, subtilisin. This led to more and more stringent testing of the peglin phage as described above and the ultimate decision that the peglin scaffold was not going to work in the phage display system. As a consequence further exploration of the truncated versions of eglin were abandoned since they were intended to serve primarily as a way to explore the binding system and identify potential binding epitopes for use with peglin. III. Exploring Search Strategies for Binders Using Randomized Peptides. The phage display system was initially developed by George Smith using randomized peptides as the source of potential binding epitopes. We began our exploration of the phage display binding system using an existing library of randomized peptides since we could begin without any recombinant DNA constructions being necessary and because that system has the most information about it already in the literature. Increasing Diversity Prior to Amplification. The initial protocol for doing phage display, taken from a phage display workshop manual, involved two rounds of panning followed by an amplification step. Adding 100 ul of a phage stock, usually at 10" PFU/ml, to a microtiter plate well for panning in the first step often resulted in only a few hundred phage at best entering the amplification step. As a consequence we modified our protocol to incorporate an amplification step after each panning step and to utilize 12 microtiter plate wells per
Page 15
sample instead of only one. This increased the number of phage entering that first most stringent amplification stepto~104PFU. We also tried using 'Immunosorb' tubes (Nunc, Naperville, IL) in that they had a much larger surface area and the expectation was that more phage would be carried through the panning cycles. In contrast to this expectation in control experiments very few phage bound. The Assorted Enzyme Set. To provide a series of targets to use to work out methodology we selected eight enzymes on the basis of commercial availability, cost and potential utility of an inhibitor. The enzymes selected were: alcohol dehydrogenase, aldolase, apha amylase, catalase, enolase, hexokinase, L-lactate dehydrogenase, and ribonuclease. Using the library of randomized peptides and six rounds of panning and amplification we found binders to four of the eight enzymes that bound with a signal of at least four times background (Table 4). The binders to ribonuclease are potentially of some interest since ribonuclease inhibitors are valuable laboratory reagents. However, no further work was done with those binders. Random Peptide Binders to Stromelysin. We screened the randomized peptide library against stromelysin using the high diversity protocol of six rounds of panning and amplification. Our expectation was that weak binders might be cleaved by the enzyme and hence would not bind. If tight binders were not cleaved and existed in the initial population that would be OK since we are primarily interested in tight binders. However, we anticipated that there might not be any tight binders and that we would need to rescue weak binders and then find ways to increase the binding affinity by further mutagenesis. To try to capture weak binders we tested as targets both native stromelysin and stromelysin inactivated by treatment with MgCl2 or CdCl2 to displace the Zn+2 required for activity in native stromelysin. Binders were found to all three forms of stromelysin (Figures 13,14,15). We sequenced thirty clones that bound to the cadmium treated stromelysin. Four different sequences were found in the thirty isolates (Figure 16). Alignments made around the amino acids common to all of the may define a stromelysin binding motif (Figure 17). We has anticipated that we would move these sequences into the full scaffold when that became available but since the full scaffold (peglin) was discovered, very late in the project, to not be active in the phage display system that was never done. IV. Production of the Target Protein Stromelysin Summary of Stromelysin Production. The objective of this project was to develop a methodology for creating protein-based inhibitors against proteins of interest. The prime target was stromelysin since that protein has been implicated in the metastasis of cancer. Being a proteinase stromelysin we expected that it would be able to accommodate a redesigned proteinase inhibitor in its active site. When we began the project there was no commercial source for this enzyme so we expected to have to build an expression system to produce it. We were able to acquire a small sample of active stromelysin and of prostomelysin at the beginning of the project from Roche Biosciences to validate our assays. After considerable effort making recombinant constructions and working out conditions to produce active stromelysin from prostromelysin we obtained a gift of a few milligrams of mature stromelysin from Parke-Davis. This was a sufficient amount for a large number
Page 16
of phage display experiments so at that time we discontinued our work on producing our own source of stromelysin. Activity Assays for Stromelysin. We obtained a gift of the mature form of stromelysin from Dr. Paul Cannon at Roche Biosciences in Palo Alto, CA. This material was diluted to a working concentration of 1 mg/ml in 25 mM Tris, pH 7.25, 10 mM CaCl2, 0.05% Brij-35 and stored at -70C in small aliquots. For our stromelysin assay we used an activity assay designed for vertebrate collagenase. That assay utilized the hydrolysis of a thiopeptide substrate, Ac-pro-leu-gly-[2-mercapto-4-methylk-pentanoyl]-leu-gly-OEt purchased from Bachern Biosicences (Philadelphia, PA) and Ellman's reagent (DNTB or 5.5'-dithio-bis(2-nitrobenzoic acid) purchased from Sigma Biochemical CO. (St. Louis, MO). The assay conditions are 50 mM MES pH 6.0, 10 mM CaCL2, 106 mM thiopeptide substrate and 1 mM DTNB in a 100 ul final volume. Thiopeptide was prepared at 76 mM in 80% acetic acid. DTNB was prepared at 20 mM in 95% EtOH. MES was prepared as a 1 M solution, filter sterilized and stored at -20C in 10 ml aliquots. CalCL2 was prepared as a 1 M solution and autoclaved. A microtiter plate reader (Molecular Devices) was used to monitor activity as indicated by change in adsorbance at 405 nm at room temperature for 30 minutes with 11 second intervals between readings. This assay has a sensitivity of about 0.5 ug/ml and a maximum of about 10-12 ug/ml (Figure 18). Activating Prostromelysin. The stromelysin enzyme is synthesized as a pro-enzyme. When we began this project it had been recently discovered that the pro sequences of several proteinases were involved in the proper folding of the enzymes. For those enzymes synthesis of the catalytic core without the pro sequence gave mostly inactive protein. It also seemed likely that expression of an active form of a proteinase in E. coli would be toxic for the bacterium. Hence we assumed that we would have to make stromelysin from prostromelysin and hence would to verify that we could convert prostomelysin to mature stromelysin. As it turns out a gene for the mature form of stromelysin does produce active enzyme and it is not prohibitively toxic in E. coli but that was not known at the time and so the following work was done. Dr. Paul Cannon also provided us with a sample of full length human prostromelysin-1 purified from IL1 stimulated human gingival fibroblast conditioned medium. The material is stored at a concentration of 0.8 mg/ml in 50 mM Tris pH 7.4, 0.2 mM NaCl, 5 mM CaC12, and 0.02% NaN3 at -70C in 100 ul aliquots. A trypsin method was used for activation using the assay described above to measure resultant stromelysin activity. Trypsin processes the 58kD prostromelysin to a 45 kD active form which partially autoprocesses to smaller active forms around 28 kD. Trypsin does not act on the substrate to produce color. Approximately 23% of the prostromelysin was activated in this experiment (Figure 19). Longer times did not increase the yield. A wider range of trypsin concentrations was tried using a gel assay to monitor conversion to a shorter form since this assays uses much less material (Figure 20). Prostromelysin was treated with 100, 25, and 4.16 uM trypsin and assayed for the production of 48 and 28 kD forms of stromelysin. Treatment with 25 uM trypsin led to conversion within 10 minutes. Treatment for 45 minutes with trypsin at any of the test concentrations led to complete degradation of the prostromelysin. The best activation treatment within the range tested was 4.16 uM trypsin for 10 minutes. Using the activity assay we found that this treatment converted approximately 9% of the prostromelysin to the active form (Figure 21).
Page 17
Clones for Prostromelysin. Several clones for prostromelysin were constructed using PCR to amplify various portions of the stromelysin-1 gene which was obtained from Dr. Lynn Matrycian. Different clones were constructed since we did not know the exact boundaries of sequence that could be expressed in E. coli and would produce a properly folded protein. Amplified DNA sequence was cloned into both pET3d and pET28a expression vectors (Novagen, Inc. Madison, WS). pET3d requires a T7 promoter for expression and hence provides very tight expression control. pET28a is driven by a T7/lac promoter and adds a his-tag to the cloned prostromelysin gene. Neither promoter is active in bacteria that do not have a source of T7 RNA polymerase. Constructions are maintained in hosts without the T7 polymerase gene since high levels of expression are usually growth inhibitory and hence lead to growth advantages for variants that have lost either the protein sequences or the expression machinery. Our plasmid constructs were transformed into BLR (DES) pLysS, a bacterial host expressing t7 RNA polymerase under tight control. Five independent transformation isolates were tested for activatable prostromelysin. Cultures were grown in LB at 37C with shaking at 250 rpm to an OD600 between 0.6 and 0.9. Expression was induced by adding IPTG to 0.4 mM for the 3d hosts and 1.0 mM for the 28a hosts and the cultures were grown for an additional 3 hrs. The cultures were chilled on ice for 5 minutes and the cells collected by pelleting at 4C. The pellets were washed in 0.25 growth volumes of 50 mM Tris pH 8.0, 10 mM CaCl2. The cells were pelleted again and frozen at -70C. To lyse the cells the pellets were thawed in a water/ice bath, resuspended in cold 1/10 growth volume 50 mM Tris, pH8.0, 10 mM CaCl2 and lysozyme was added to 0.2 mg/ml. After 20 minutes on ice the material was sonicated at 90%intermediate output (Branson sonifier) for 1 minute. This lysate was cleared by centrifugation in an Eppendorf micocentrifuge. The supernatant was collected and stored at 4C. Total protein was determined using the Pierce BCA method. Two of the five clones tested, strom3 and strom9, produced stromelysin activity after activation with trypsin (Figure 22). Purification of his-tagged prostromelysin. We chose to work up the strom9 clone since it has a histag to facilitate purification while the strom3 clone does not. Duplicate cultures of bacteria containing the strom9 expresser plasmid were grown up and induced as described above. Samples were collected every hour for 6 hours after induction. Lysates were prepared as described above and analyzed by electrophoresis on polyacrylamide gels (Figure 23). On induction a band at 33 kD is seen to increase in intensity with time of induction. The predicted size for this prostromelysin product is 32 kD as the clones were intentionally truncated during construction to make the shortest possible 'pro' form of stromelysin. To purify the soluble his-tagged prostromelysin from these cells a frozen pellet from 1250 ml of culture was thawed on ice and resuspended in 10 ml of dHOH and the final solution made up to 0.2mg/ml in lysozyme, 10 ug/ml in DNase, 10 mM CaCl2, and 1 mM PMSF. After 30 minutes on ice the solution was sonicated (5-30 seconds bursts with 30 seconds of cooling between each). The solution was then cleared by centrifugation at 4C. The supernatant was filtered through a 0.45 micron syringe filter and kept on ice while a 4 ml his-bind resin (Novagen, Inc. Madison, WS) column was prepared. Resin was washed with 12 ml of dHOH and the wash removed after centrifugation at 300 g. The column was then charged with nickel by suspending the resin for 30 minutes in 20 ml of 50 mM NiS04 and then removing the charging solution by centrifugation. Finally the column was washed with 12 ml of 5 mM imidazole, 500 mM NaCl, 20 mM Tris pH 7.9. Lysate and resin were Page 18
then mixed and placed on a shaker which gently agitated (50 rpm) the mix at 4C for 1 hour. The mixture was then centrifuged and the supernatant removed for analysis. The resin and bound his-tagged prostromelysin was then washed five times for 10 minutes with 10 ml of 60 mM imidazole, 500 mM NaCl, 20 mM Tris pH 7.9 at 4C on the shaker. The final mixture was transferred to a poly-prep chromatography column (BioRad, Hercules, CA) and the final wash collected as flow through. Prostromelysin was eluted using 14 ml of 1 M imidizole, 500 mM NaCl, 20 mM Tris pH 7.9. 1 ml fractions were stored at 4C for analysis on 12% polyacrylamide-1% SDS gels. The bulk of the prostromelysin eluted in the second two elution fractions (Figure 24). To determine the yield of protein from the column, aliquots from elution fractions were analyzed on an electrophoresis gel alongside of a series of lanes with different amounts of BSA. By estimating which BSA lane had a band of equal intensity of the prostromelysin bands we could estimate the amount of protein in the prostromelysin fractions using much less material than a traditional protein assay. From this analysis (Figure 25) we estimate that we recovered 2 BSA equivalent milligrams of prostromelysin per liter of culture. External Source for Stromelysin. After considerable time and effort in making our own stromelysin by activating prostromelysin and getting fairly low yields we decided to invest in attempts to make the mature form of stromelysin directly from a clone. We constructed expression vector clones containing only the catalytic domain of stromelysin. Thirteen of fourteen isolates tested made active stromelysin. During this period we had been communicating with Dr. Quezang Ye at Parke-Davis about his work with catalytic domain clones and purification methods for stromelysin. We were also interested in getting their clone for stromelysin. After some haggling by the lawyers, Parke-Davis agreed to provide us with stromelysin for the remainder of the project instead of the clone. We were subsequently sent 6.7 mg of stromelysin. V. Characterization of the scaffold protein (eglin c) As part of our effort to convert eglin c into a useful inhibitor of other proteins we have carried out a series of investigations into the properties of eglin c as a protein. Those investigations have yielded a method for the quantitative assessment of hypotheses concerning the determinants of protein structure. This method is an extension of the use of mutagenesis and combinatorial libraries to study proteins and we believe the new method has broad utility. We think that this is an important new approach for the study of proteins and are very excited about this development. We anticipate that it will be the focus of research in my laboratory for the next decade. The method is described in the accompanying manuscript which has been submitted to the Proceedings of the National Academy of Science USA (August 30, 1999). The work in this manuscript and the two others that have come from this project are summarized in the sections below. A Method for the Quantitative Assessment of Hypotheses Concerning the Determinants of Protein Structure (manuscript 3, submitted). Suppose that one had a hypothesis concerning the determinants of some aspect of structure in a particular protein. An approach in use since our discovery of a method for making directed mutations in DNA is to mutagenize the residues involved and determine what happens to the protein. A later adaptation of this approach is to randomize the residues involved in a combinatorial library and assess the consequences both in terms of the fraction of variants in the library that pass whatever tests are applied and in terms of characterization of variants that pass and fail the tests. A more hypothesis oriented approach was Page 19
developed in the laboratory of Dr. Michael Hecht in which he constructed combinatorial libraries in which each library member conformed to a hypothesis about the hydrophobic or hydrophilic nature of the residue in a particular position in the protein. Each library variant had the same hydrophilic-hydrophobic pattern at the test sites in the protein. The large fraction of the library variants that passed the tests was then used to assert that the pattern was sufficient to encode structure and that the specific details of the residues were not critical. We have extended this approach by the addition of two features; one modifies the library construction approach and the other changes the mode of analysis of the library. First, we use resin-splitting technology (29,30) to facilitate the construction of arbitrarily complex libraries that are free of the constraints imposed by the genetic code. In all of the previous studies the libraries that were made were all rendered degenerate either by randomizing residues or by using degenerate codons (codons with mixed nucleotides at one or more sites). This limits the nature of the libraries that can be constructed. Split-resin technology can be used to synthesize an arbitrarily complex set of degenerate oligonucleotides and frees library construction from the previous constraints. The second feature that we have added to the assessment of libraries involves the use of regression analysis to extend the analytical power of combinatorial library experiments. Appropriate selection of the nature of variants in such a library makes it possible to use regression analysis for quantitative assessment of specific hypotheses and, by averaging over the effects of many factors, to extract accurate information regarding partial effects contributing to protein structure formation. Regression analysis can also be used to assess several competing hypotheses using a single library, in contrast to the approach using the fraction of the library variants that remains active as the metric. Regression analysis provides access to new information by providing a formalism for the quantitative evaluation of the consequences of the effects defined in a hypothesis and a statistical assessment of the degree to which variant behavior can be attributed to them. With this approach we have shown that we can independently assess and reproduce a known determinant of protein structure, a-helix propensities. Helix propensities represent the contribution to the free energy of denaturation made by the various amino acids in solvent exposed positions in an a-helix. Regression parameters derived from the analysis of a 455 member sample from a library wherein four solvent-exposed sites in an a-helix can contain any of nine different amino acids are highly correlated (P < 0.0001, R2 > 0.97) to the relative helix propensities for those amino acids, as estimated by a variety of biophysical and computational techniques. This agreement encourages us to believe that our approach can provide quantitative assessments of other hypotheses about the relations between amino acid sequence, stability and structure. Effect of a Poly-His-Terminal Extension on Eglin c Stability (manuscript #1; Anal. Biochem 1998). To facilitate protein purification, a poly-his-terminal extension has been incorporated onto eglin c. In this manuscript we reported that the his-tag incorporation does not effect eglin c stability. Thermal denaturations monitored by circular dichroism spectropolarimetry showed that the free energy of denaturation did not change upon his-tag incorporation. Nonideality and Protein Thermal Denaturation (manuscript #2; Biopolymers 1999). We studied the thermal denaturation of eglin c by suing CD spectropolarimetry and differential scanning calorimetry (DSC). At low protein concentrations, denaturation is consistent with the classical two-state model. At Page 20
concentrations greater than several hundred uM, however, the calorimetric enthalpy and the midpoint transition temperature increase with increasing protein concentration. These observations suggested the presence of intermediates and/or native state aggregation. However, the transitions are symmetric, suggesting that intermediates are absent, the DSC data do not fit models that include aggregation, and analytical ultracentrifugation (AUC) data show that native eglin c is monomeric. Instead, the AUC data show that eglin c solutions are nonideal. Analysis of the data gives a second virial coefficient that is close to values calculated from theory and the DSC data are consistent with the behavior expected from nonideal solutions. We conclude that the concentration dependence is caused by differential nonideality of the native and denatured states. This nonideality is hypothesized to arise from the high charge of the protein at acid pH and is exacerbated by the low buffer conditions in which these experiments are traditionally carried out.. Our conclusions may explain differences between van't Hoff and calorimetric denaturation enthalpies observed for other proteins whose behavior is otherwise consistent with the classical two-state model.
Page 21
KEY RESEARCH ACCOMPLISHMENTS • • •
we have discovered that eglin c is not a good scaffold protein for phage display we have discovered that a circularly permuted version of eglin (peglin) also does not function well as a scaffold protein in phage display we have discovered a new way make and analyze combinatorial libraries to assess hypotheses concerning the determinants of protein structure
Page 22
REPORTABLE OUTCOMES Manuscripts Published 1. Waldner, JC, Lahr, SJ, Edgell, MH, and Pielak, GJ. Effect of a polyhistidine terminal extension on eglin c stability. Anal. Biochem. 263 116-118 (1998) 2. Waldner, JC, Lahr, SJ, Edgell, MH, and Pielak, GJ. Nonideality and protein thermal denaturation. Biopolymers 49 471-479 (1999) Submitted 3. Lahr, SJ, Broadwater, A, Carter, CW, Jr., Collier, ML, Hensley, L, Waldner, JC, Pielak, GJ, and Edgell, MH. Patterned library analysis: a method for the quantitative assessment of hypotheses concerning the determinants of protein structure. Proc. Nat. Acads.Sci. USA (submitted 1999) Patents none Degrees Mr. Steven Lahr is writing his PhD thesis. The research that will be reported in that thesis is the biophysical characterization of eglin c and the development of our method for analyzing combinatorial libraries which we are calling patterned library analysis. Cell lines none Informatics none Funding Applied for I applied to NIH in 1998 for funding to explore and apply the patterned library analysis method. That proposal was not funded but the review indicated that they liked the approach but felt the method needed the assessment of publication. I intend to submit an amended proposal for the November 1 NIH deadline. Employment opportunities Mr. Lahr has secured a post-doctoral position with Dr. William DeGrado and will join that laboratory in October 1999.
Page 23
CONCLUSIONS Although we were not able to develop an appropriate scaffold protein that could be used to generate high affinity inhibitors for proteins involved in the metastasis of cancer I feel that the general concept for making inhibitors is still viable. This is based on the fact that we were able to accomplish all of the parts of the project except for the very crucial part of developing an appropriate scaffold protein. A different scaffold protein or a more extensively engineered version of eglin might be expected to work. We had anticipated that the difficult part of the project would be to recover weak binders from the target proteinase since weak binders might be expected to be substrates, to be cleaved and hence to not bind. However, we were able to find binders to stromelysin or papain in the first pass of panning with a phage display library containing a site cleavable by the target. What else have we learned about making protein based inhibitors and stromelysin? 1. Prostromelysin can be made in large quantities in E. coli. Most of the product is in a soluble form, that is, not inclusion bodies. The protein is not processed to a toxic form in the bacterial cell. 2. In vitro activation of prostromelysin with trypsin is very inefficient. The best we could ever do was convert about 10% of prostromelysin to catalytically active stromelysin. 3. The catalytic portion of the stromelysin protein when produced in E. coli is active. That is, the protein does not need the pro-sequence to fold correctly. 4. We were able to find peptide binders to more than half of the 'randomly' chosen enzymes in our test panel. If those binding sites are in the active site of those enzymes then it should be possible to construct protein based inhibitors for a very wide range of enzymes. What else have we learned? As part of our efforts to characterize the binding epitope display framework protein, eglin c, we discovered a new approach to assess hypotheses concerning the determinants of protein structure. This approach is based on two realizations. The first is that one can build combinatorial libraries encoding proteins that are free of the constraints of the genetic code by using an existing technology called resin splitting to synthesize the desired degenerate oligonucleotides. The second is that an increased level of quantitative information could be extracted from combinatorial libraries using regression analysis. Employing these insights we carried out a proof-of-principle experiment using the new method in which we were able to reproduce known values for ct-helix propensity indicating that the method does indeed work. Mutagenesis is the major tool for exploring hypotheses concerning the determinants of protein structure. Any advance in our capacity to use this tool should have broad utility. We are very hopeful that the approach that we have discovered is such an advance.
Page 24
PERSONNEL SUPPORTED (In full or in part) Anne Broadwater (Technician) Martha L. Collier (Technician) Dr. Marshall Hall Edgell (Principal investigator) Lucinda Hensley (Technician) Steven J. Lahr (Graudate student) Paula Davis-Searles (Graduate student) Dr. Gary Pielak (co-Principal investigator) Jennifer C. Waldner (Graduate student) PERSONNEL WHO WORKED ON THE PROJECT WITHOUT SUPPORT Jill Sherman (Undergraduate student) Sharon Smith (Master's student, Carolina Central University) Amy Turner (Undergraduate student)
Page 25
Binding Epitope Region C-Terminus
N-Terminus minus seven disorganized residues
To circularly permute -^ open here
FIGURE 1. Ribbon Diagram of Eglin C. Note that attaching a phage particle to the C-terminus of eglin c might block access to the binding epitope. Our construction of a circularly permuted eglin removes the seven disorganized residues from the N-terminus, adds a four residue tight turn to connect the N- and C-terminal ends, and opens the protein to create new N- and C-termini at the point indicated in the figure.
Page 26
eglin
peglin
Figure 2. E coli lysates expressing pegin or eglin c. Lane Lane Lane Lane
1. 2. 3. 4.
Extract from Extract from Extract from Extract from
BLR BLR BLR BLR
with with with with
a a a a
gene gene gene gene
for for for for
15% Polyacrylamide gel with 1% SDS.
Page 27
PEGLIN 0 hr. after induction PEGLIN 1.5 hr. after induction PEGLIN 3.0 hr. after induction EGLIN C 3.0 hr. after induction
Peglin binding 0.8
Peglin 1.1 vs SC O
Peglin 4.1 vs SC
"-0-"
SA291A1 vsSC
0.6-
ITS
o
o o 0.4-
0.2
^0™=
—I 10+10
1— 10+9
10+8
10+7
10+6
10+5
10+4
10+3
Phage Concentration (PFU/ml) Figure 3. Binding to Wells Coated with Subtilisin. The two peglin constructs both bind to the native target for eglin c, subtilisin, more readily than does a phage displaying a peptide which binds to streptavidin. Since all proteins bind to some extent to this proteinase we wanted to determine relative binding efficencies. Binding at high phage concentrations is due to non-specific binding. Page 28
|—
EGLIN Primer w/ Ear site
~^ Degenerate Primer w/ Ear site
[— LacZ alpha
j
mBAX PLASMID DNA
Figure 4. Earl strategy for library construction. PCR can be used to amplify an entire plasmid DNA. If one of the primers is degenerate (the unfilled box represents the degenerate region) then the amplified products will contain that degeneracy. By using sites on the ends of the primers (bars show sites) such as Earl or EamI that cut outside of their recognition sites one can produce amplified product that when cleaved and ligated contains no restriction site sequence. That is, one can cut and ligate the product DNA at chosen sites that are totally independent of the distribution of sites in the template DNA as long as there are not additional Earl sites in the template.
Page 29
[-
|— LacZ alpha
EGLIN
1
Primer w/ Earn site
M Primer w/ Earn site
mBAX PLASMID DNA
Figure 5. Construction of a 'White' mBAX Phage. A -200 bp deletion in the alpha fragment of beta-galactosidase within the mBAX plasmid was created by using PCR and primers cotaining Earn restriction sites. Restriction enzymes recognizing these sites cut outside of the recognition site and hence allow one to cut and splice anywhere one likes within a sequence indepedent of the distribution of restriction sites in the target DNA. This presumes that there are not additional Earn sites in the target DNA. Circular plasmid DNA was used as the template for PCR.
Page 30
6 A
Subtilisin standard curve (2 folds)
1/
B C
M13
1.0X1014PFU
D
peglin 01
1.2x10HPFU
E
peglin02
2.0 x 1013 PFU
F
peglin 1.1 < 1.0x108 PFU
G
peglin 2.1
1.2x1010 PFU
H
Two-fold serial dilutions
Figure 6. Inhibitory Effects of Various Phage preparations without Dilution (i.e. similiar amounts of PEG). Aliquots of PEG concentrated phage stocks were mixed with 8.7 x 10$ molecules of subtilisin for 60 minutes. Substrate was then added and cleavage of substrate followed by color development at 405 nm. Reduced color development indicates inhibition. The fact that each phage preparation shows essentially the same amount of inhibition (slopes in column 2 or 3) independent of the number of phage inthe well suggests that the inhibition is not due to phage. Wells Gl and G3 artifactually show no color development (presumably no substrate got added).
Page 31
MOLECULAR DEVICES Raw Data (Plate) DATA FILE DESCRIPTION PROTOCOL DESCRIPTION MODE WAVELENGTH MEAN TEMP
DATA 9/24 11_14_47 9/24/98mlcPeglinclones1.1-5.2vsSubtlisin&catalase plateA PRINTED: 9/24/98 Phage: 18 hr ON -30 min color development Endpoint AUTOMIX: ON 405 CALIBRATION: ON 24.90°C SET TEMP: OFF
Optical Density
1
2
3
4
5
6
7
8
9
10
1 1
12
A
0.064
0.068
0.074
0.074
0.092
0.081
0.103
0.078
0.133
0.080
0.126
0.088
B
0.068
0.067
0.072
0.072
0.087
0.072
0.102
0.085
0.118
0.081
0.138
0.091
C
0.064
0.060
0.071
0.072
0.085
0.069
0.085
0.068
0.107
0.075
0.121
0.078
D
0.065
0.062
0.074
0.069
0.085
0.083
0.087
0.068
0.111
0.072
0.113
0.090
E
0.168
0.137
0.125
0.133
0.133
0.127
0.140
0.124
0.190
0.136
0.200
0.161
F
0.100
0.118
0.110
0.114
0.127
0.125
0.137
0.121
0.173
0.135
0.219
0.143
G
0.085
0.091
0.092
0.099
0.114
0.103
0.116
0.105
0.156
0.110
0.165
0.113
H
0.092 . 0.090
0.090
0.097
0.107
0.098
0.113
0.098
0.135
0.102
0.158
' 0.110
Figure 7a. Binding (or lack of binding) of Peglin to Stromelysin. Rows A through D are wells coated with subtilisin. Rows E through H are coated with catalase to which the peglin clones should not bind. Odd columns have 1:5 dilutions of a phage stock and even rows have 1:50 dilutions. The different dilutions allow one to assess phage concentration dependent binding. Columns 1,2 contain peglin isolate vl.l Columns 3,4 isolate vl.2 Columns 5,6 isolate v2.1 Columns 7,8 isolate v2.2 Columns 9,10 isolate v3.1 Columns 11,12 isolate v3.2; A phage preparation containing phage that bound to its cognate target (subtilisin) would have larger numbers in rows A-D than rows D-H.
Page 32
MOLECULAR DEVICES Raw Data (Plate) DATA FILE DESCRIPTION PROTOCOL DESCRIPTION MODE WAVELENGTH MEAN TEMP
DATA 9/24 11j0_45 9/24/98mlcPeglinclones1.1-5.2vsSubtlisin&catalase-plateB PRINTED: 9/24/98 Phage: 18 hr ON -30 min color development Endpoint AUTOMIX: ON 405 CALIBRATION: ON 24.80°C SET TEMP: OFF
Optical Density 1
2
3
4
5
6
7
8
9
10
1 1
12
A
0.064
0.062
0.075
0.067
0.081
0.063
0.066
0.062
0.059
0.085
0.356
0.078
B
0.060
0.054
0.077
0.059
0.075
0.060
0.068
0.061
0.061
0.057
0.285
0.081
C
0.063
0.056
0.093
0.058
0.074
0.058
0.063
0.063
0.063
0.060
0.301
0.084
D
0.069
0.058
0.082
0.062
0.082
0.063
0.069
0.064
0.066
0.071
0.367
0.095
E
0.157
0.106
0.359
0.183
0.131
0.121
0.111
0.110
0.149
0.123
0.482
0.208
F
0.152
0.100
0.369
0.192
0.126
0.111
0.120
0.112
0.150
0.122
0.489
0.248
G
0.177
0.109
0.373
0.195
0.141
0.119
0.121
0.112
0.152
0.128
1.909
2.015
H
0.178
0.115
0.412
0.198
0.141
0.124
0.119
0:117
0.161
0.151
1.635
2.019
Figure 7b. Binding (or lack of binding) of Peglin to Stromelysin. Rows A through D are wells coated with subtilisin. Rows E through H are coated with catalase to which the peglin clones should not bind. Odd columns have 1:5 dilutions of a phage stock and even rows have 1:50 dilutions. The different dilutions allow one to assess phage concentration dependent binding. Columns 1,2 contain peglin isolate v4.1 Columns 3,4 isolate .v4.2 Columns 5,6 isolate v5.1 Columns 7,8 isolate v5.2; Wells Gl 1,12 and HI 1,12 were loaded with a catalase binding phage to serve as our phage binding controls. A phage preparation containing phage that bound to its cognate target (subtilisin) would have larger numbers in rows A-D than rows D-H.
Page 33
Eglin
Figure 8. Conversion of Egin to Teglin. The ten amino acid loop containing the binding epitope and the underlying beta strands containing residues that contribute to two framework to loop salt-bridges is removed and joined via a cysteine bond. The teglin structure would not of course maintain the beta sheet conformation. The closed loop structure is fused to the Ml3 pill protein via a QGGGG linker peptide.
Page 34
thr leu
pheala.
Teglin
Figure 9. Conversion of Tegin to Teglin-papain. The two resdiues in teglin on each side of what would be the cleavage site in a substrate (indicated by the arrow) were replaced in teglin-papain with the residues in a papain cleavage site.
Page 35
Xhol site TCCTCCCJCGAG TGC GGT ACC ATC NNS TTC GCT NNS NNS NNS ATC CYS GLY THR ILE ??? PHE ALA ??? ??? ??? ILE ™
Papain cleavage site
Xbal site GAC CGC ACC CGT TCC TTC TGT TAGGGIG_GCGGTGGCTCTAGATCCTCC ASP ARG IHR ARG SER PHE CYS
««««„««^»^A-r^-r«^Ä^ ACCGCCACCGAGATCTAGGAGG ■d
ReversePrimer
Figure 10. Construction of a Teglin Based Papain Cleavge Sequence Library. An ologonucleotide was syntehsized with three randomized codons just downstream of the papain binding site and one randomized codon just upstream of the cleavage site. The oligonucleotide was rendered double stranded using the primer shown. The double stranded DNA was then cleaved with Xho and Xba and inserted into the Ml3 display phage containing the teglin gene such that the native subtilisin binding eptiope is replaced by the degenerate sequence.
Page 36
BINDERS TO PAPAIN 1.9-
-1.9
1.8-
-1.8
1.7-
-1.7
1.6-
-1.6 D
1.5 D D
1.4
D D
1.3 H
D n u D □□□D
an ÜB-
o o
n
D
0.7
-1.3
CÜ
1.2 D
D
-1.1
D
-
D
D
D
D D
D D
D
1
-0.9
D
0.90.8-
-1.4 %D
1.11- □
an
EP
1.5
r
D D
1.2
o
a
°D
-0.8
DD
D
D D
0.6-
-0.7 -0.6
0.5-
-0.5
0.4-
-0.4
0.3-
-0.3
0.2-
0.2
0.1-
hO.l 0
0
INDIVIDUAL ISOLATES, CLONE NUMBER Figure 11. ELISA Values for Individual Clones After Six Rounds of Panning Against Papain. Individual clones are isolated at random from the population which was present after six rounds of panning against papain and bound to a well in a 96-well plate coated with papain. The amount of phage that binds is determined by an ELISA assay using a primary antibody against wild-type Ml3 bacteriophage. Page 37
Design TGC GGT ACC ATC MNJS TTC GCT NNS NNS NNS ATC GAC CGC ACC CGT TCC TTC TGT cys gly thr ile xxx phe ala xxx xxx xxx ile asp arg thr arg ser phe cys
Higher Affinity Papain Binders TGG GTA CCA TCG GGT TCG CTG GGA CGC GGG ATC GAC CGC ACT TGT TCC TTC TGT trp val pro ser gly ser leu gly arg gly ile asp arg thr cys ser phe cys TGG CGG TAC CAT CAA GTT CGC TCG GAG GCT CAT CGA CCG CAC CCG TTC CTT CTT trp arg tyr his gin val arg ser glu ala asp arg pro his pro phe leu leu TGC GGT ACC ATC GGG TTC GCT CCG AGG CTG ATC GAC CGC ACC CAT TCC TTC TTT cys gly thr ile gly phe ala pro arg leu ile asp arg thr his ser phe phe TGG CGG TAC CAT CAC GTT CGC TCC GAG CCC GAT CGA CCG CAC CCG TTC CTT CTG trp arg tyr his his val arg ser glu pro asp arg pro his pro phe leu leu TGC GGT ACC ATC GAC TTC GCT AAG AGG ACG ATC TAC CGC ACC CAT TCC TTC TGG cys gly thr ile asp phe ala lys arg thr ile tyr arg thr his ser phe trp
Intermediate Affinity Papain Binders TAC CAT CTA GTT CGC TGG GGG GAG GAT CGA CCG CAC CCG TTC CTT CTG TCG GGT tyr his leu val arg trp gly glu asp arg pro his pro phe leu ser gly gly TGC GGT ACC ATC GAG TTC GCT GGG GGC GGG ATC GAC CGC ACC CGT TCC TTC TGT cys gly thr ile glu phe ala gly gly gly ile asp arg thr arg ser phe cys TGC GGT ACC ATC TGG TTC GCT GGG GGG GAT ATC GAC CGC ACC CGT TCC TTC TGT cys gly thr ile trp phe ala gly gly asp ile asp arg thr arg ser phe cys
Figure 12. Sequences of Binders to Papain. All but two of the sequences have undergone some mutation relative to the library design which has removed the possbility of forming a constrained loop through a disulphide bond.
Page 38
BINDERS TO UNTREATED STROMELYSIN
D
0.8- D
a
D D
D
D
D
D
D
D
D
0.6
D
d? D
OD405
0oo
n Gn n
D
DQD
ODD
D D
D EU
0.4-
D
□
^ D
n
D
□ D
0.2-
□ □ D
D □
D
DD G
J
D
D
1
25
50
75
100
Well # Figure 13. ELISA Values for Individual Clones After Six Rounds of Panning Against Stromelysin. Individual clones are isolated at random from the population which was present after six rounds of panning against stromelysin and bound to a well in a 96-well plate coated with stromelysin. The amount of phage that binds is determined by an ELISA assay using a primary antibody against wild-type Ml3 bacteriophage. Page 39
BINDERS TO STROMELYSIN TREATED WITH MgCl2
D D D
0.8-
D
DD
0.6-
□
JD ^P
DnDD
%
D
R]
na^D^nn^
tf°
Dl
D
□°
J
CE|L
n
D
n
D U
n
D
D
a
D
OD405 0.4-
0.2-
D
□
D D
-T-
25
50
75
D D
100
Well# Figure 14. ELISA Values for Individual Clones After Six Rounds of Panning Against Stromelysin Treated with MgCl2. Individual clones are isolated at random from the population which was present after six rounds of panning against MgCl2 treated stromelysin and bound to a well in a 96-well plate coated with stromelysin. The amount of phage that binds is determined by an ELISA assay using a primary antibody against wild-type Ml3 bacteriophage. Page 40
BINDERS TO STROMELYSIN TREATED WTIH CdCl2 D D
%□
D
D
D
□
nDD °
D 0.8-
d? D
□
D
D
D
D
°
DD . °n° na 1
D
D
Cr
D
D
D
rn □ CD
°DD
D D
SQ
D
D D
D
0.6
OD 405
0.4-
0.2-
D
D D
□ D
D
D
D
dP
D
nD
1
25
50
Well#
D D
D
~~r~ 75
100
Figure 15. ELISA Values for Individual Clones After Six Rounds of Panning Against Stromelysin Treated with CdCI2. Individual clones are isolated at random from the population which was present after six rounds of panning against CdCl2 treated stromelysin and bound to a well in a 96-well plate coated with stromelysin. The amount of phage that binds is determined by an ELISA assay using a primary antibody against wild-type Ml3 bacteriophage. Page 41
Design (27)
TCC TCG AGT NNK NNK NNK NNK NNK NNK NNK NNK NNK NNK NMK NNK TCT AGA CCT TCC TCG AGT CCG CTT GAG AGG TTG ATG GCG CGG ATG GOT ACT CCT TCT AGA CCT pro leu glu arg leu met ala arg met ala thr pro
(1)
TCC TCG AGT CGG TCT GGG TTG GAG TCT TAT TGG AGG AGT GCG GAG TCT AGA CCT arg ser gly leu glu ser tyr trp arg ser ala glu
(1)
TCC TCG AGT TTG GAT GCG TGG CCG GAT GGT CCG AAG CGG ATT GCG TCT AGA CCT leu asp ala trp pro asp gly pro lys arg ile ala
(1)
TCC TCG AGT GGT AGG TCG GCT TGG ACG ATT GAT GGG ACT GTT GTG TCT AGA CCT gly arg ser ala trp thr ile asp gly thr val val
Figure 16. Sequences of Clones Which Bound to CdCl2 Treated Stromelysin. Most of the isolates had a single sequence. The number of isolates with a given sequence are indicated in parenthesis.
Page 42
met met ser asp ile ser
ala ala ala ala ala ala
arg thr glu trp ser trp
met pro ser pro arg thr
glu ala ser trp lys gly
arg arg arg arg arg arg
leu met ser ser ile ser
met ala gly ala ala ala
Figure 17. Alignments of Sequences from Stromelysin Binding Clones. Only alanine and arginine are present in the variant portion of all of the isolates. Alignments are presented around each of these two residues.
Page 43
Vmax
12.5
ug/ml Stromelysin
Figure 18. Standard Curve for Stromelysin. A colormetric microtiter plate assay was used to monitor amounts of stromelysin. A kinetic curve of color development using as substrate, Ac-Pro-Leu-Gly-[2-mercapto-4-methyl-pentanoyl]-Leu-Gly-0Et, was collected in wells of a microtiter plate at 405 nm in a Molecular Devices plate reader. Vmax was calculated and used as the metric for amounts of stromelysin. Reactions were at room temperature for 10 minutes with readings taken every 11 seconds.
Page 44
2345678910
] Lag Time = |0:00r
I 1
12
Vrnax Pts. = [if 0/164
Figure 19. Trypsin Activation of Prostromelysin. Samples of mature stromelysin and trypsin treated prostromelysin assayed with the thiopeptide substrate. Each box shows the color production as a function of time in wells of a microtiter plate. Readings were taken every 11 seconds. Reactions were carried out in 50 mM Tris pH 7.4, 5 mM CaCl2 and 200 raM NaCl. Trypsin cleavage was for 30 minutes at 37C. Rows A-C Mature Stromelysin column 1: 5.0 ug/ml stromelysin column 2: 2.5 ug/ml stromelysin column 3: 1.2 ug/ml stromelysin column 4: 0.6 ug/ml stromelysin Rows D,E Prostromelysin treated with 25 uM trypsin column 1: 100 ug/ml prostromelysin column 2: 50 ug/ml prostromelysin column 3: 25 ug/ml prostromelysin colimn4 12 ug/ml prostromelysin Row F Prostromelysin treated with 75 uM trypsin column 1: 100 ug/ml prostromelysin column 2: 50 ug/ml prostromelysin column 3: 25 ug/ml prostromelysin colimn4 12 ug/ml prostromelysin Page 45
>• i%- ' 12
3 4
5
7
8
9 10 11
12 13.14 15
I
2
3 4
5
6 7
8
9
10 II
IP 13 14 I0
60kDa. 50kDa . 40kDa. *
60kDa. 50kDa
30kDa.
40kDa—A
20kDa.
W
30kDa_||
20kDa
1 10kDa Marker 2 Prostromelysin 3 Prostrom. 10min. 4.16^M trypsin 4 Prostrom. 20min. 4.16(xM trypsin 5 Prostrom. 30min. 4.16|.iM trypsin 6 Prostrom. 10min. 4.16(.iM trypsin +l 7 Prostrom. 10min. 2pM trypsin 8 Prostrom. 20min. 2(.iM trypsin 9 Prostrom. 30min. 2pM trypsin 10 Prostrom. 60min . 2^iM trypsin 11 Prostrom. 10min . 1|aM trypsin 12 Prostrom. 20min . 1|iM trypsin 13 Prostrom. 30min . 1 [xM trypsin 14 Prostrom. 50min . 1|aM trypsin 15 Prostrom. 60min . 1(iM trypsin
1 10kDa Marker 2 Prostromelysin 3 Prostrom. 10min. 100|iMtrypsin 4 Prostrom. 20min. IOO^IM trypsin 5 Prostrom. 30min. IOO^IM trypsin 6 Prostrom. 10min. 25(iM trypsin 7 Prostrom. 20min. 25^iM trypsin 8 Prostrom. 30min. 25|.iM trypsin 9 Prostrom. 10min. 4.16(.iM trypsin 10 Prostrom. 20min. 4.16(iM trypsin 11 Prostrom. 30min. 4.16|.iM trypsin 12 Prostrom. 45min. 100|iM trypsin 13 Prostrom. 45min. 25^iM trypsin 14 Prostrom. 45min. 4.16|.iM trypsin 15 10kDa Marker
Figure 20. Gel Assay for Trypsin Activation of Prostromelysin. Prostromelysin has a molecular weight of 58 kD. Mature stromelysin is 45 kD and a processed active form at 28 kD. Digestions were done in 50 mM Tris pH 7.4, 5 mM caCl2, 200 mM NaCl,
Page 46
1
2
3
4
5
6
7
8
9
10
1 1
12
A B C D
73g.
E F G H OD Limit =10.100
I Lag Time = !0:00
j Vmax Pts. = no/i64
!
Figure 21. Trypsin Activation of Prostromelysin at 4.16 uM Trypsin. A colormetric assay for the amount of active stromelysin was used to monitor the activation of prostromelysin. Each box represents the time course of color development for one well in a microtiter plate. The substrate is Ac-Pro-Leu-Gly-[2mercapto-4-methyl-pentanoyl]-Leu-Gly-0Et. Reactions were at room temperature for 30 minutes. The OD represented by the box height is 0.1. Row A. Mature form of stromelysin Column 1: 5.0 ug/ml stromelysin Column 2: 2.5 ug/ml stromelysin Column 3: 1.2 ug/ml stromelysin Column 4: 0.6 ug/ml stromelysin Column 5: 0.3 ug/ml stromelysin Row B. Prostromelysin untreated Column 1: 48 ug/ml prostromelysin Column 2: 24 ug/ml prostromelysin Column 3:12 ug/ml prostromelysin Column 4: 6 ug/ml prostromelysin Column 5: 3 ug/ml prostromelysin RowC. Prostromelysin treated for 30 minutes at 37C with 4.16 uM trypsin Column 1: 48 ug/ml prostromelysin Column 2: 24 ug/ml prostromelysin Column 3:12 ug/ml prostromelysin Column 4: 6 ug/ml prostromelysin Column 5: 3 ug/ml prostromelysin Page 47
B
r } } }
D
s*-^
"1
H ^
'p
A
A
A
